Skip to main content

mlx_fast_scaled_dot_product_attention

Function mlx_fast_scaled_dot_product_attention 

Source
pub unsafe extern "C" fn mlx_fast_scaled_dot_product_attention(
    res: *mut mlx_array,
    queries: mlx_array,
    keys: mlx_array,
    values: mlx_array,
    scale: f32,
    mask_mode: *const c_char,
    mask_arr: mlx_array,
    sinks: mlx_array,
    force_fused: bool,
    s: mlx_stream,
) -> c_int